Vision & Robotics 2013 Fair

J.A.I.M.S. continuously develops new vision techniques for a various of applications.
These techniques contain not only image analyzing or vision software but also new image creating techniques e.g. illumination and optics. The range of applications are very divers and contain the whole area from food industry until high tech industry, from large to very small products and from Ultra Violet until Mid Wave Infra Red.
Every year at the Vision & Robotics Fair, J.A.I.M.S. shows new vision techniques, new vision solutions and new vision projects.

This year J.A.I.M.S. demonstrates the following topics at the stand 42.

White Cabbage stump detection. (Dutch: Witte Kool stam detectie)

Witte Kool stam detection For automatic processing of white cabbage it is necessary to know the position of the stump.
First step in this process is detection of the stump followed by rotating the cabbage in the right position so a fast cutting process can cut of a small piece of the stump. This is necessary because the stump had a slightly rotting surface. This step is followed by cutting the cabbage in small slices.
The shown image shows the color (left top) and grey (right top) image of the cabbage. The red cross in the grey images shows the center of the detected stump. On the left bottom the processed image is shown, the white area is the detected stump.
On the right bottom the normalized intensity profile of the grey image (red line) and the results of the processed image (white line). The intensity line is taken at the position of the horizontal pixel line (see grey image). The contrast of the stump in the processed image is much better than the contrast in the grey image, so a reliable detection after processing the image is possible. Actually no detection of the stump is possible using the grey image.

White Cabbage stump and stem detection. (Dutch: Witte Kool stam en stengel detectie)

Witte Kool stam and stem detection With this technique it is possible to detect also the stem of the leaves as shown in the images. The color images (left top) and the grey images (right top) are visible.
On the left bottom the processed image for stump detection is shown and on the right bottom the processed image for the stem detection is shown.
Actually it is also possible to detect the green part of the cabbage leafs.

White Cabbage rot spot detection. (Dutch: Witte Kool rotte plek detectie)

Witte Kool rot detection Very interesting of this technique is also the possibility of detecting the rot spots of the cabbage.
Here two settings are shown, on the left bottom the processed image, the very dark and very decayed spots and area are detected. On the right bottom the processed image of a light decayed area is shown.

Bacon and fat detection.(Dutch: speklap)

Bacon detection In these images the difference between fat and meat is shown. On the left bottom the processed image of fat and on the right bottom the processed image of the meat of the bacon is shown.
In this situation it is possible using basic vision algorithms to detect the amount of fat and or the amount of meat.

Pork steak.(Dutch: hamlap)

Pork steak detection This technique is also useful for detection of a small quality difference between meat.
In this case the meat quality detection is based on the color difference between a good and a lower quality piece of meat.
The lower quality is still within the quality limits. And of course with this very good contrast between both qualities it is possible to do the necessary measurements on the steak image.

Nanometer accurate position measurement.

Nano precision measurement Of course it is not possible to visualize a feature of a few nanometer by a normal optical vision system. But it is possible to measure the position of a larger feature of several microns with an accuracy of a few nanometers.
J.A.I.M.S. shows the accurate position measurement of a two magnetic conductors within an accuracy of 5 nanometers (3 times standard deviation). The set up as shown is lab prototype setup using a normal objective (20x) and a 5 mega-pixel camera. The dimension of the measured magnetic conductors is 200x3,5 micrometers. The field of view is about 250x200micrometer.
Also shown the distance between the middle lines of both conductors.
This lab prototype is followed by a more stabilized setup using a 20/0,35 Long Working Distance objective. The position of an other product with a feature of 20x3 micron is measured. This feature is also measured in a large part of the image area. Measurement are done using a 3D piezo actuator.
Software is special developed for a maximum of accuracy for this type of feature. Unfortunately this set up is not shown.

Surface quality inspection of a 5 mm lens on micron scale.

The production of high volume high quality lenses are done by so called injection moulding process.
And of course every injection moulding process needs an injection mould, and the quality of the product is as good as this mould or less. To create a high accurate injection mould special tooling is required and used to create the special lens shape in the mould.
Lens_Quality water line visible Lens_Quality water line NOT visible Lens_Quality scratch visible Lens_Quality scratch NOT visible
Still it can happen that the mould and the resulting lenses do have some damages or imperfections. It is difficult to visualize these defects on micron scale using common optics and common illuminations.
Therefore J.A.I.M.S. developed:
  • special optics to create a good as possible image
  • special illumination to visualize the defects and imperfections on micron scale of this curved lens surface.
  • .
    The image (basically 3602x3602 pixels) shows the tooling grooves and the defects.
    First image shows a so called "water line" on the top of the lens.
    On the second image regular illumination is used, this "water line" is not visible!
    The third image a scratch (at the right bottom) is visible. The large "disturbance" is a marker made by a waterproof marker.
    The fourth image, again a regular illumination is used, the scratch is not visible!!
    It is also important to see the tooling grooves and the defects so we are certain the optics are focused on the lens surface. These tooling grooves does not really disturbe the quality of the lens.

    Quality inspection of the xenon burner.

    The xenon lamp, used in modern cars, is a precision product.
    To guarantee the accuracy and life time of this lamp each burner is measured and inspected. J.A.I.M.S. shows the first test setup to measure:
  • the burner shape,diameter w.r.t. the rotation angle
  • The excentricity of the burner balloon
  • the electrode distance.
  • the electrode displacement w.r.t. the balloon
  • the width of the squeezed quarts glass, width compression joint
  • Xenon Burner Xenon Burner animated
    These measurements are done by rotating the burner.
    Basically more measuerements are done but not implemented in this software version.
    These measurements are implemented on the production machine but it's obvious that this production equipment can not be shown on the Vision & Robotics fair.

    J.A.I.M.S. shows a lot of images as a results of other projects:

  • Lens surface Quality Inspection of 1 mm lenses.
    These high volume high quality lenses, built up of two separated lenses with special shaped surfaces, have to be inspected. Scratches, defects and other imperfections are not allowed.
    Inspection of these lenses are done manually. The required inspection quality is inspector depended so a vision system should improve the inspection quality and saving costs. To solve this problem microscopic optics and special illumination, created by J.A.I.M.S., is used to visualize the defects.
    Result is a very good homogeneous illuminated image with a very good defect contrast. A reliable detection and measurement of the defect can be performed using basic vision software algorithms.

  • Impeller Inspection Impeller surface quality inspection. Very small surface defects are detected of this complex product with smoothed surface.

  • Adjustment of the xenon burner in 6D using a 6 motor driven Hexapod.

  • Adjustment a lithographically master with respect to a previous applied pattern on a screen. Adjustment is done in in 6D using a 6 motor driven Hexapod.

  • Vision controlled gluing of a scintillator and a UV sensor device for medical purposes.

  • Glass quality control of Ultra High Pressure burners.

  • Glass quality control of High Pressure burners.

  • Glass quality of wine glass and glass tubes.

  • Measurement of and adjusting the Röntgen collimator for medical purposes.

  • Measurement of the radius and imperfections of a diamond cutting tool with a radius of 350 micrometer.

  • Quality measurement of cement for street lamp production.

  • Inspection and measurement of the wire loop of transistors.

  • Product identification reading of laser applied code on glass.

  • Glass fiber diameter measurement.

  • Etc. etc...

  • Lens Surface Quality Thooth-error Glass_fiber_test FIT photo equipment_1 small.JPG Schotel D 3D small UHP burner 01

    Damaged cement lamp holder glue top view Xenon_lamp burning Product Identification Number Wire Loop Typical Example HPI Quality Inspection Diamont Tool Measurement